As we see in the demo.py, it is a tool for single pictures.
So if I want to process videos as you demonstrated in the paper, how can I do this?
I split the video into photos, then process them in batches, and finally connect the photos back into a video. However, this approach presents two problems: one is the continuity of the video, and the other is the misalignment of the model in the connected video, making it seem shaky. Therefore, I'd like to know how to resolve this issue.
As we see in the demo.py, it is a tool for single pictures.
So if I want to process videos as you demonstrated in the paper, how can I do this?
I split the video into photos, then process them in batches, and finally connect the photos back into a video. However, this approach presents two problems: one is the continuity of the video, and the other is the misalignment of the model in the connected video, making it seem shaky. Therefore, I'd like to know how to resolve this issue.